Enterprise Intelligence Group
Phantom inventory in retail: what to check before buying software
Investigate stock shown in your system but unavailable to sell. Use a practical diagnostic to check inventory records, data handoffs and replenishment decisions.
Your system says a product is in stock. A colleague cannot find it, an order cannot be fulfilled, or the shelf stays empty. Before changing the forecast or buying a detection tool, establish what the stock record actually represents—and what happened to the units it describes.
Phantom inventory is recorded stock that is not physically available. It is one form of inventory-record inaccuracy. An empty shelf alone does not prove it: the item may be in the backroom, reserved, damaged, misplaced or recorded under a different status. Those situations require different actions.
Start with the item, location and time
Choose a specific item and location. Preserve the system balance and its timestamp before a correction changes the evidence. Compare it with a physical check made on a consistent basis, accounting for sales and movements between the two times.
Record which measure you are comparing: physical on-hand, sellable stock, available-to-promise or another defined balance. Check whether reservations, damaged units and in-transit stock are included. Two systems can disagree because their definitions differ, not because either calculation is wrong.
Treat an unexplained discrepancy as a question to investigate—not permission for an automatic stock adjustment.
Follow the stock movement through the systems
Review the transactions relevant to the selected case: receipts, sales, returns, transfers, damage and adjustments. Check their sequence, identifiers and units of measure. A correctly transmitted message may still have been rejected by the receiving application or applied to the wrong item or location.
Ask whether a case-pack quantity has been interpreted as individual units, whether a returned item became sellable before inspection, or whether a transfer left one location without reaching the other location’s record. These are diagnostic possibilities, not assumptions about your operation.
Check the replenishment decision separately
A corrected record does not necessarily produce an order. Review the rules and constraints that applied at the decision time: minimum quantities, pack sizes, ordering schedules, supplier availability and any human override.
Equally, a forecast adjustment will not repair a receipt that was posted incorrectly. Keep the demand estimate, stock balance and ordering decision visible as separate parts of the investigation. The purpose is to find the break in the process before choosing the remedy.
Use a short evidence record
| Review question | Evidence to capture | Decision it supports |
|---|---|---|
| What balance are we comparing? | Item/location identifiers, balance definition and timestamp | Whether the comparison is valid |
| What is physically available? | Shelf/backroom check, condition and stock status | Whether the problem is missing stock, accessibility or eligibility for sale |
| Which movements explain the balance? | Relevant transaction history, quantities and sequence | Where to investigate the discrepancy |
| Did the systems agree? | Source/destination records, mappings and rejected transactions | Whether a data handoff needs correction |
| Why did replenishment act—or not act? | Parameters, constraints and override record at the decision time | Whether the ordering workflow also needs attention |
| Who can approve the next action? | Named owner, evidence reviewed and approval boundary | A controlled correction and follow-up |
Keep the original evidence with the resolution. A reason code is more useful when it points to a supported explanation rather than becoming a convenient label for an unknown cause.
An illustrative case—not a client result
A store record shows six units and a shelf check finds none. One possible explanation is that the units are in the backroom. Another is that a receipt was posted for six units but the delivery was short. A third is that the units are present but not eligible for sale.
The same visible symptom therefore does not justify the same correction. Confirm the physical position, trace the relevant movements and involve the authorized owner before changing the record. Then check whether the ordering process responds as intended. The example illustrates a review method; it does not report an EIG engagement or measured financial improvement.
Where research and detection tools can help
ECR Retail Loss’s June2026 research examines how inventory-record errors can be predicted and how checks can be prioritized. It also distinguishes modeled economic value from results that still require retailer-specific validation. Read the research.
For a software evaluation, ask what evidence the tool uses, when that evidence was available, and how alerts are validated. Agree how exceptions reach the responsible team and which actions require approval. Test against your operating conditions; a convincing demonstration is not a substitute for an agreed acceptance test.
Decide what kind of help is needed
If the investigation points to ordering rules, exceptions or disconnected planning decisions, explore EIG’s forecasting and replenishment consulting.
If it points to item definitions, data harmonization, mappings or system handoffs, explore retail data and integration analysis.
EIG’s role is to help define the problem, organize the evidence and evaluate the changes required. This guide is not an automated detection service or an instruction to change production stock balances. Start with a defined operating question; agree the scope of any review before sharing business data.
The EIG Library · Insights · AI Maturity Assessment · Contact